LDP/LDP/guide/docbook/abs-guide/str-test.sh

#!/bin/bash
#  str-test.sh: Testing null strings and unquoted strings,
#+ but not strings and sealing wax, not to mention cabbages and kings . . .

# Using   if [ ... ]

# If a string has not been initialized, it has no defined value.
# This state is called "null" (not the same as zero!).

if [ -n $string1 ]    # string1 has not been declared or initialized.
then
  echo "String \"string1\" is not null."
else  
  echo "String \"string1\" is null."
fi                    # Wrong result.
# Shows $string1 as not null, although it was not initialized.

echo

# Let's try it again.

if [ -n "$string1" ]  # This time, $string1 is quoted.
then
  echo "String \"string1\" is not null."
else  
  echo "String \"string1\" is null."
fi                    # Quote strings within test brackets!

echo

if [ $string1 ]       # This time, $string1 stands naked.
then
  echo "String \"string1\" is not null."
else  
  echo "String \"string1\" is null."
fi                    # This works fine.
# The [ ... ] test operator alone detects whether the string is null.
# However it is good practice to quote it (if [ "$string1" ]).
#
# As Stephane Chazelas points out,
#    if [ $string1 ]    has one argument, "]"
#    if [ "$string1" ]  has two arguments, the empty "$string1" and "]" 


echo


string1=initialized

if [ $string1 ]       # Again, $string1 stands unquoted.
then
  echo "String \"string1\" is not null."
else  
  echo "String \"string1\" is null."
fi                    # Again, gives correct result.
# Still, it is better to quote it ("$string1"), because . . .


string1="a = b"

if [ $string1 ]       # Again, $string1 stands unquoted.
then
  echo "String \"string1\" is not null."
else  
  echo "String \"string1\" is null."
fi                    # Not quoting "$string1" now gives wrong result!

exit 0   # Thank you, also, Florian Wisser, for the "heads-up".